An Improved Scheme for Interest Mining Based on a Reconfiguration of the Peer-to-Peer Overlay

Tan et al. proposed a scheme to improve the quality of a file search in unstructured Peer-to-Peer systems by focusing on the similarity of interest of the participating peers. Although it certainly improves the cost/performance ratio of a simple flooding-based scheme used in conventional systems, the Tan's method has a serious drawback such that a query cannot reach a target peer if a requesting peer is not connected with the target peer through a path consisting of peers to have similar interest to the given query. In order to overcome such drawback of the Tan's method, we propose a scheme to reconfigure the underlying network in such a way that a requesting peer has a neighbor interested in the given query, before transmitting a query to its neighbors. The performance of the proposed scheme is evaluated by simulation. The result of simulation indicates that it certainly overcomes the drawback of the Tan's method.